A NEW METHOD FOR THE CONSTRUCTION OF RELATEDNESS SEMANTIC NET. BASED ON DISTRIBUTIONAL SEMANTICS

Abstract

In this paper, proposal of local semantic net. created by a new special algorithm  called automatic semantic net algorithm (ASN). finding the relevance degree between words in self-document without background knowledge is the focus of attention of this proposal, depends mainly on the characteristics of distributional semantic which contributed to build a specific word-net for each document. In addition to that taking into account the convergence between the sequence of words in document addition to frequency of each word. used the amendment on (Wu & Palmer) similarity metric measurer to make it commensurate with the proposed semantic net. experimental results it was very convincing compared to prior knowledge semantic net approach and the modify on standard similarity metric measure gave better results.